For 1-4/7, 6/7:
Reviewed-by: Roland Scheidegger srol...@vmware.com
I think though using two exclusive booleans is really ugly. Probably
should just use a tgsi_return_type enum instead indeed, and treat the
unorm/snorm cases as float. I'm not entirely sure what the unorm/snorm
types
Am 11.06.2015 um 17:33 schrieb Rob Clark:
On Thu, Jun 11, 2015 at 10:53 AM, Roland Scheidegger srol...@vmware.com
wrote:
For 1-4/7, 6/7:
Reviewed-by: Roland Scheidegger srol...@vmware.com
I think though using two exclusive booleans is really ugly. Probably
should just use
and a separate patch)
BR,
-R
On Wed, Jun 10, 2015 at 2:55 PM, Roland Scheidegger srol...@vmware.com
wrote:
My biggest problem with that is the initial case I saw as a problem:
draw is going to modify these shaders in some cases (aaline stage for
example), adding its own sampler
...@gmail.com wrote:
On Tue, Jun 9, 2015 at 9:32 AM, Roland Scheidegger srol...@vmware.com
wrote:
Am 09.06.2015 um 15:00 schrieb Rob Clark:
On Tue, Jun 9, 2015 at 5:01 AM, Jose Fonseca jfons...@vmware.com wrote:
On 09/06/15 04:03, Rob Clark wrote:
On Mon, Jun 8, 2015 at 10:50 PM, Roland
long line...
}
static void close_prog(struct program *p)
Reviewed-by: Roland Scheidegger srol...@vmware.com
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
that is unrelated and a separate patch)
BR,
-R
On Wed, Jun 10, 2015 at 2:55 PM, Roland Scheidegger srol...@vmware.com
wrote:
My biggest problem with that is the initial case I saw as a problem:
draw is going to modify these shaders in some cases (aaline stage for
example), adding its own sampler
.
Reviewed-by: Roland Scheidegger srol...@vmware.com
Am 16.06.2015 um 12:29 schrieb Jose Fonseca:
It's not totally clear whether other Mesa drivers can safely cope with
over-sized UBOs, but at least for llvmpipe receiving a UBO larger than
its limit causes problems, as it won't fit into its
No. And as far as I know noone is working on it for llvmpipe. If I'd
write a to do list for llvmpipe, that would probably be on page 50 or so
:-). But patches always welcome...
Roland
Am 17.06.2015 um 14:21 schrieb Ilyes Gouta:
Hi,
On Wed, Jun 17, 2015 at 1:03 PM, Dragomir Ivanov
, and with dynamic indexing
this isn't known at compile time as it can differ for each texture
bound), but everything GL4 is not even really on a to do list... Feel
free to work on it though...
Roland
Am 17.06.2015 um 15:00 schrieb Ilyes Gouta:
Hi,
On Wed, Jun 17, 2015 at 1:51 PM, Roland
Am 12.06.2015 um 22:38 schrieb Rob Clark:
On Fri, Jun 12, 2015 at 9:47 AM, Rob Clark robdcl...@gmail.com wrote:
On Thu, Jun 11, 2015 at 8:36 PM, Roland Scheidegger srol...@vmware.com
wrote:
Am 12.06.2015 um 02:02 schrieb Rob Clark:
On Thu, Jun 11, 2015 at 5:47 PM, Brian Paul bri
Am 12.06.2015 um 02:02 schrieb Rob Clark:
On Thu, Jun 11, 2015 at 5:47 PM, Brian Paul bri...@vmware.com wrote:
On 06/11/2015 02:38 PM, Rob Clark wrote:
From: Rob Clark robcl...@freedesktop.org
Some hardware needs to know the sampler type. Update the blit related
shaders to include SVIEW
);
+ total_verts_per_buffer *
shader-num_invocations);
This can't get very large, right?
Otherwise,
Reviewed-by: Roland Scheidegger srol...@vmware.com
debug_assert(output_verts-verts);
#if 0
@@ -592,7 +598,7 @@ int draw_geometry_shader_run(struct draw_geometry_shader
(mach, inst, offsets);
Reviewed-by: Roland Scheidegger srol...@vmware.com
I think though there's a lot more required to make it work in softpipe...
Roland
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman
Looks ok, it's massively simpler and shouldn't break as often.
Reviewed-by: Roland Scheidegger srol...@vmware.com
Am 28.05.2015 um 17:57 schrieb Jose Fonseca:
It doesn't do everything we want. In particular it doesn't allow to
detect jumps or return opcodes. Currently we detect the x86's RET
simplified these things in llvmpipe at some point but never did for
softpipe (which is mostly why the actual arithmetic is different for the
llvmpipe soa sampling path vs. softpipe). I might tackle it one day...
Thanks!
Reviewed-by: Roland Scheidegger srol...@vmware.com
Also fix the gather lod calculation
Am 27.05.2015 um 13:47 schrieb Ilia Mirkin:
On Wed, May 27, 2015 at 7:42 AM, Roland Scheidegger srol...@vmware.com
wrote:
Am 27.05.2015 um 09:45 schrieb Dave Airlie:
From: Dave Airlie airl...@redhat.com
This is just for softpipe, llvmpipe won't work without
some changes.
Signed-off
Yes, that looks better to me. It may be a bit inconsistent to only do
that for inputs/outputs and not for temp but at least this way there's
no additional limit on the number of arrays.
Roland
Am 26.05.2015 um 13:04 schrieb Marek Olšák:
From: Marek Olšák marek.ol...@amd.com
v2: use less
Thanks for addressing this, some minor issues.
Am 27.05.2015 um 03:14 schrieb Dave Airlie:
Roland pointed out my previous attempt was lacking, so I enhanced the
texwrap piglit test, and tested them. This fixes the offset calculations
in a number of areas by adding the offset first, it also
Am 27.05.2015 um 09:45 schrieb Dave Airlie:
We need indexed queries to retrieve the geom shader info.
Signed-off-by: Dave Airlie airl...@redhat.com
---
src/gallium/drivers/softpipe/sp_context.h | 2 +-
src/gallium/drivers/softpipe/sp_prim_vbuf.c | 4 ++--
probably inherent to how multiple streams work.
So, other than the things I mentioned, for the series:
Reviewed-by: Roland Scheidegger srol...@vmware.com
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo
I use this sometimes for debugging things in llvmpipe, so I don't think
that's a good option. Without it getting the actual assembly of the
shaders would be quite annoying.
It is unfortunately true that (due to the heavy use of the unstable C++
API) it breaks very often, but as long as it isn't
,
true);
+ LLVMAddTargetDependentFunctionAttr(func,
no-frame-pointer-elim-non-leaf, true);
+#endif
+
LLVMRunFunctionPassManager(gallivm-passmgr, func);
func = LLVMGetNextFunction(func);
}
Reviewed-by: Roland Scheidegger srol...@vmware.com
/05/15 12:56, Roland Scheidegger wrote:
I use this sometimes for debugging things in llvmpipe, so I don't think
that's a good option. Without it getting the actual assembly of the
shaders would be quite annoying.
It is unfortunately true that (due to the heavy use of the unstable C++
API
Don't worry about the AoS stuff. Only meant to do simple things.
Looks good overall, I guess it makes sense to not split execution too
(so you'd have native hw vector size there), llvm should handle that
pretty well these days (the sse intrinsics won't get used that way
probably (though there's a
Am 05.07.2015 um 15:47 schrieb Ilia Mirkin:
On Sun, Jul 5, 2015 at 9:25 AM, Marek Olšák mar...@gmail.com wrote:
From: Marek Olšák marek.ol...@amd.com
Other approaches are being considered:
1) Don't use resource wrappers (views) and pass all view parameters
(format, layer range, level) to
Am 06.07.2015 um 19:54 schrieb Ilia Mirkin:
That's right. Except really what might have happend was
occl query;
write X;
more drawing;
write X+1;
and then on the CPU, you see X+1. So the tests are always for = X.
And if you have more than 2^32 submits, you cry, because I'm *sure*
that
Do we really need this?
I think ideally we wouldn't really need any limit on this, as it is
already limited by the max number of the temps (though obviously that
number is higher). d3d10 doesn't seem to have any such limit neither.
If some driver can't do it, I think it's not unreasonable to
, Apr 4, 2015 at 10:50 AM, srol...@vmware.com wrote:
From: Roland Scheidegger srol...@vmware.com
llvm goes crazy when doing that, using way more memory and time, though
there's
probably more to it - this points to a very much similar issue as fixed in
8a9f5ecdb116d0449d63f7b94efbfa8b205d826f
This doesn't really do what the commit message claims - it just adds a
cap bit, not actual support for arb_cull_distance (which was already
there), so the log should be changed accordingly.
Apart from what was already mentioned (that is, the cap bit isn't used
by st/mesa which instead mistakenly
Overall, this looks like a good idea to me. Probably needs some changes
in our internal code but the idea is quite sane.
Roland
Am 24.05.2015 um 13:19 schrieb Marek Olšák:
Hi,
The reason I add this is that TGSI doesn't allow indirect indexing of inputs
and outputs. Consider this:
MOV
Am 24.05.2015 um 13:19 schrieb Marek Olšák:
From: Marek Olšák marek.ol...@amd.com
---
src/gallium/auxiliary/tgsi/tgsi_scan.c | 11 +--
src/gallium/auxiliary/tgsi/tgsi_scan.h | 2 ++
2 files changed, 11 insertions(+), 2 deletions(-)
diff --git
This case wasn't dealing with floats, though.
That said I guess it's a pointless discussion. If the compiler doesn't
think it's a good idea I'll just add the parentheses just to shut it up...
Roland
Am 25.05.2015 um 21:19 schrieb Axel Davy:
If one side is NaN, then
!(x == y) is true, whereas x
I can't see what's wrong with the code.
I thought not using parantheses there is quite common, is that some new
warning from some super duper new compiler version?
Roland
Am 25.05.2015 um 02:46 schrieb Roger Brown:
From mesa-10.5.6 compilation.
Is this an oversight ?
lp_query.c:
should probably be *4*4...
oops.
Yeah that works much better :) Thanks for the hint!
On Mon, May 25, 2015 at 1:42 PM, Roland Scheidegger srol...@vmware.com
wrote:
The only thing I can think of is an illegal setup of the constant
buffer, because the index was not verified before for direct
On 05/19/2015 04:46 PM, Adam Jackson wrote:
On Tue, 2015-05-12 at 23:34 +0200, Roland Scheidegger wrote:
I like faster!
I don't really understand though where the wins come from.
softpipe can't have much pending work in the first place (maybe tile
cache?), llvmpipe potentially has
Am 01.08.2015 um 01:26 schrieb Matt Turner:
gcc actually generates this for us now that we use -fno-math-errno
(which is weird, since lrintf()/lrint() don't set errno) but clang still
does not. Presumably helps MSVC as well.
Reduced .text size by 8.5k with gcc before -fno-math-errno.
Am 01.08.2015 um 02:44 schrieb Matt Turner:
On Fri, Jul 31, 2015 at 4:41 PM, Ilia Mirkin imir...@alum.mit.edu wrote:
On Fri, Jul 31, 2015 at 7:26 PM, Matt Turner matts...@gmail.com wrote:
gcc actually generates this for us now that we use -fno-math-errno
(which is weird, since lrintf()/lrint()
(double x)
Is there any need for this one? In any case,
Reviewed-by: Roland Scheidegger srol...@vmware.com
+{
+ return lrint(x);
+}
+
+#endif
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo
Am 01.08.2015 um 03:02 schrieb Matt Turner:
On Fri, Jul 31, 2015 at 5:50 PM, Roland Scheidegger srol...@vmware.com
wrote:
Am 01.08.2015 um 01:26 schrieb Matt Turner:
gcc actually generates this for us now that we use -fno-math-errno
(which is weird, since lrintf()/lrint() don't set errno
I have some doubts of this. Given that IROUND was just plain incorrect,
I think we should just use _mesa_lroundevenf() and see what happens.
lroundf() is quite the shocker, my libm's implementation is totalling 40
instructions (!) for it (not counting the last return, vs 1 for lrintf).
Now it's
FWIW d3d10 has the same requirements, the any filter needs to be
supported for half float formats, but is optional for float formats
(d3d10.1 made it required for floats).
https://msdn.microsoft.com/en-us/library/windows/desktop/cc627090%28v=vs.85%29.aspx
So there might well be some more hw where
Am 08.08.2015 um 22:45 schrieb Jason Ekstrand:
Mesa formats and gallium formats are defined a bit differently. In mesa
there are packed formats which are based on byte-order within a 8, 16, or
32-bit word and there are array formats which are simply an array of 8,
16, or 32-bit values. In
Am 08.08.2015 um 23:26 schrieb Jason Ekstrand:
On Sat, Aug 8, 2015 at 2:14 PM, Roland Scheidegger srol...@vmware.com wrote:
Am 08.08.2015 um 22:45 schrieb Jason Ekstrand:
Mesa formats and gallium formats are defined a bit differently. In mesa
there are packed formats which are based on byte
Makes sense.
Reviewed-by: Roland Scheidegger srol...@vmware.com
Am 12.08.2015 um 12:45 schrieb Marek Olšák:
From: Marek Olšák marek.ol...@amd.com
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=74329
v2: add a CAP for half floats
drivers should not expose the CAPs
Ping?
Am 09.08.2015 um 17:28 schrieb srol...@vmware.com:
From: Roland Scheidegger srol...@vmware.com
I used this as some testing ground for investigating some compiler
bits initially (e.g. lrint calls etc.), figured I could do much better
in the end just for fun...
This is mathematically
;
+ dbl_type.width *= 2;
+ lp_build_context_init(bld.bld_base.dbl_bld, gallivm, dbl_type);
+ }
bld.mask = mask;
bld.inputs = inputs;
bld.outputs = outputs;
Looks good to me. I'm not entirely happy with the build_gather, but it
is fixable later, your choice.
Reviewed-by: Roland
Am 30.06.2015 um 03:42 schrieb Dave Airlie:
On 30 June 2015 at 09:36, Roland Scheidegger srol...@vmware.com wrote:
Am 29.06.2015 um 22:18 schrieb Dave Airlie:
On 30 June 2015 at 00:58, Roland Scheidegger srol...@vmware.com wrote:
Don't worry about the AoS stuff. Only meant to do simple things
Am 29.06.2015 um 22:18 schrieb Dave Airlie:
On 30 June 2015 at 00:58, Roland Scheidegger srol...@vmware.com wrote:
Don't worry about the AoS stuff. Only meant to do simple things.
Looks good overall, I guess it makes sense to not split execution too
(so you'd have native hw vector size
CC mesa-dev.
This looks good to me. I am starting to wonder though why we don't just
use lrintf() and let the compiler sort it out (for x86 too).
Though actually some quick experiments show that:
- llvm's clang will always use libm lrintf call. Which then will do
(x86_64) cvtss2si %xmm0,%rax as
+#else
return lrint(x);
+#endif
}
#endif
Reviewed-by: Roland Scheidegger srol...@vmware.com
I'm wondering though if this is really exactly implementing lrintf why
we're using a different name and not just something like _mesa_lrintf().
Though arguably _mesa_roundeven is the same
I don't think that's quite true in general.
For gpus which have combined ds buffers I can'see why you'd wanted to do
separate clears for depth and stencil in this case (i.e. doing
pipe-clear for depth, then draw a quad for clearing stencil).
At least for simple hw like llvmpipe which don't have
Actually, since the code says it can only happen with a non-full stencil
mask, isn't clearing depth/stencil with a non-full stencil mask
incredibly rare?
Roland
Am 31.07.2015 um 18:51 schrieb Roland Scheidegger:
I don't think that's quite true in general.
For gpus which have combined ds
On Fri, Jul 31, 2015 at 7:44 PM, Roland Scheidegger srol...@vmware.com
wrote:
Actually, since the code says it can only happen with a non-full stencil
mask, isn't clearing depth/stencil with a non-full stencil mask
incredibly rare?
Roland
Am 31.07.2015 um 18:51 schrieb Roland Scheidegger:
I
thought this was hit more often, but apparently
not. Nevermind.
Marek
On Fri, Jul 31, 2015 at 7:44 PM, Roland Scheidegger srol...@vmware.com
wrote:
Actually, since the code says it can only happen with a non-full stencil
mask, isn't clearing depth/stencil with a non-full stencil mask
Am 31.07.2015 um 18:44 schrieb Matt Turner:
On Fri, Jul 31, 2015 at 7:13 AM, Roland Scheidegger srol...@vmware.com
wrote:
CC mesa-dev.
This looks good to me. I am starting to wonder though why we don't just
use lrintf() and let the compiler sort it out (for x86 too).
Though actually some
# http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43052
I'm wondering if we should add that to the scons build as well (both of
these, of course).
Reviewed-by: Roland Scheidegger srol...@vmware.com
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http
Hmm I guess we can't convert it to something else when we don't have a
matching MESA_FORMAT?
Roland
Am 30.07.2015 um 03:48 schrieb Dave Airlie:
From: Dave Airlie airl...@redhat.com
GL3.3 added GL_ARB_texture_rgb10_a2ui, which specifies
a lot more things than just rgb10/a2ui.
While
I guess the question is if those depth bounds floats should be in the
dsa state or set separately. That would mostly depend on if you expect
frequent changes to the values or not, at least that's the pattern
usually followed. Either way though looks ok to me.
Reviewed-by: Roland Scheidegger srol
- 1, dst_bits - 1);
- else
+ else if (src_bits dst_bits)
return x (src_bits - dst_bits);
+ else
+ return x;
}
static inline unsigned
Reviewed-by: Roland Scheidegger srol...@vmware.com
(I don't care either way for the snorm case).
Roland
Am 09.08.2015 um 18:47 schrieb Matt Turner:
On Sun, Aug 9, 2015 at 3:57 AM, Jose Fonseca jfons...@vmware.com wrote:
As currently only GCC x86_64 builds where using it.
---
src/util/rounding.h | 16 +---
1 file changed, 13 insertions(+), 3 deletions(-)
diff --git
LONG_BIT
#endif
#else
return lrint(x);
For the series:
Reviewed-by: Roland Scheidegger srol...@vmware.co
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
Forgot to mention, strictly speaking only __SSE__ is necessary for
_mesa_lroundevenf, so it would work on these shiny Pentium 3 and Athlon
XP... The double version (though it's unused) however requires __SSE2__.
Roland
Am 09.08.2015 um 19:23 schrieb Roland Scheidegger:
Am 09.08.2015 um 18:47
Looks good to me. Would have been much less trouble if we'd just
returned an int, to hell with lrintf's long return type no callers will
care about the long returned anyway...
Reviewed-by: Roland Scheidegger srol...@vmware.com
Am 09.08.2015 um 23:40 schrieb Jose Fonseca:
More portable. Based
Am 09.08.2015 um 20:08 schrieb Jose Fonseca:
On 09/08/15 17:47, Matt Turner wrote:
On Sun, Aug 9, 2015 at 3:57 AM, Jose Fonseca jfons...@vmware.com wrote:
Neither MSVC nor MinGW defines LONG_BIT. For MSVC this was not a
problem as
it doesn't define __x86_64__ macro (it's GCC specific.)
I guess though you'd need these bits when implementing things like
ARB_fragment_shader_ordering? (Thus stuff actually looks useful but I
don't know if anybody wants to implement it in the near term.)
Roland
Am 14.08.2015 um 18:27 schrieb Francisco Jerez:
The hardware documentation relating to
I'm hardly a build expert, but looks good to me.
Reviewed-by: Roland Scheidegger srol...@vmware.com
Am 15.08.2015 um 00:22 schrieb Vinson Lee:
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=91591
Signed-off-by: Vinson Lee v...@freedesktop.org
---
scons/gallium.py | 1 +
1 file
Am 09.08.2015 um 12:11 schrieb Oded Gabbay:
On Sun, Aug 9, 2015 at 2:43 AM, Jason Ekstrand ja...@jlekstrand.net wrote:
On Sat, Aug 8, 2015 at 3:14 PM, Oded Gabbay oded.gab...@gmail.com wrote:
On Sun, Aug 9, 2015 at 12:26 AM, Jason Ekstrand ja...@jlekstrand.net
wrote:
If I understand your
Am 10.08.2015 um 07:00 schrieb Jason Ekstrand:
On Sun, Aug 9, 2015 at 6:46 AM, Roland Scheidegger srol...@vmware.com wrote:
Am 09.08.2015 um 12:11 schrieb Oded Gabbay:
On Sun, Aug 9, 2015 at 2:43 AM, Jason Ekstrand ja...@jlekstrand.net wrote:
On Sat, Aug 8, 2015 at 3:14 PM, Oded Gabbay
Could mention this is actually a gl spec bug in the commit message, but
either way:
Reviewed-by: Roland Scheidegger srol...@vmware.com
Am 21.07.2015 um 22:55 schrieb Brian Paul:
Since s3tc works for cube maps and 2D arrays, it should also work for
cube arrays. NVIDIA's driver supports
implementation. The only
case where resource_copy_region must be used is copying compressed
textures.
Marek
On Mon, Jul 20, 2015 at 9:56 PM, Roland Scheidegger srol...@vmware.com
wrote:
Theoretically resource_copy_region should be the right function for
that. Both ARB_copy_image
, and for the rest of the series,
Reviewed-by: Roland Scheidegger srol...@vmware.com
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
Well they are not sharable by definition (per-context entities), it is
not meant to work. So, it definitely should be fixed one way or another.
Roland
Am 23.07.2015 um 17:44 schrieb Marek Olšák:
Hi Brian,
Like the commit message says, I have no idea how Firefox does it. My
commit only
Am 17.07.2015 um 16:18 schrieb Brian Paul:
On 07/16/2015 03:15 PM, srol...@vmware.com wrote:
From: Roland Scheidegger srol...@vmware.com
In particular, we were incorrectly accepting s3tc (and lots of others)
for CompressedTexSubImage3D (but not CompressedTexImage3D) calls with 3d
targets
gallium is already using fabsf in lots of other places. Supported by
older msvc versions.
Roland
Am 14.07.2015 um 11:50 schrieb Iago Toral:
This one will need wrapping for fabsf.
Reviewed-by: Iago Toral Quiroga ito...@igalia.com
On Mon, 2015-07-13 at 16:22 -0700, Matt Turner wrote:
---
Am 13.07.2015 um 17:58 schrieb Emil Velikov:
On 11 July 2015 at 19:39, srol...@vmware.com wrote:
@@ -186,17 +172,24 @@ static inline void emit_cb_setup(struct r100_context
*r100,
/* XXX others? BE/LE? */
Drop the BE/LE part of the comment ?
switch (mesa_format) {
+/*
Am 16.07.2015 um 02:18 schrieb Ilia Mirkin:
On Wed, Jul 15, 2015 at 7:40 PM, Brian Paul bri...@vmware.com wrote:
This adds the new glGetTextureSubImage() and
glGetCompressedTextureSubImage() functions. Also update the
dispatch sanity test program.
v2: remove stray brace, move xi:include
Am 14.07.2015 um 04:03 schrieb Matt Turner:
On Mon, Jul 13, 2015 at 5:13 PM, Roland Scheidegger srol...@vmware.com
wrote:
Did you replace 2 of them by exp2f but one by exp2f on purpose?
Yes, two of them are converting their results to floats so they were
converted to exp2f().
Upon
Am 15.07.2015 um 21:44 schrieb Matt Turner:
On Mon, Jul 13, 2015 at 5:13 PM, Roland Scheidegger srol...@vmware.com
wrote:
Did you replace 2 of them by exp2f but one by exp2f on purpose?
I don't think we can use exp2/exp2f in gallium. This requires msvc 2013
(all of exp2, exp2f, powf are c99
Am 15.07.2015 um 21:58 schrieb Matt Turner:
On Wed, Jul 15, 2015 at 12:44 PM, Matt Turner matts...@gmail.com wrote:
On Mon, Jul 13, 2015 at 5:13 PM, Roland Scheidegger srol...@vmware.com
wrote:
Did you replace 2 of them by exp2f but one by exp2f on purpose?
I don't think we can use exp2
Am 17.07.2015 um 16:18 schrieb Brian Paul:
On 07/16/2015 03:15 PM, srol...@vmware.com wrote:
From: Roland Scheidegger srol...@vmware.com
In particular, we were incorrectly accepting s3tc (and lots of others)
for CompressedTexSubImage3D (but not CompressedTexImage3D) calls with 3d
targets
Did you replace 2 of them by exp2f but one by exp2f on purpose?
I don't think we can use exp2/exp2f in gallium. This requires msvc 2013
(all of exp2, exp2f, powf are c99, powf is supported by older msvc but
the others are not). I guess though could throw some wrappers into
c99_math.h.
Roland
Am
Am 12.07.2015 um 19:30 schrieb Ilia Mirkin:
I asked this on IRC, but figured I'd get wider distribution for the
question. The situation is that nv50 doesn't support RGBA32 MS8
textures. However QuerySamplesForFormat will try its hardest to find
supported sample counts, which means it'll go
solution will require
creating a TargetMachine during gallivm initialization and pulling the
DataLayout from there. This will be a somewhat invasive change, and it
will need to be validatated on multiple LLVM versions.
Reviewed-by: Roland Scheidegger srol...@vmware.com
---
src/gallium
Theoretically resource_copy_region should be the right function for
that. Both ARB_copy_image and d3d10 ResourceCopyRegion define this as
essentially the equivalent of memcpy.
I guess the difficulty is that GL's view classes are a bit different
compared to d3d10 typeless groups - d3d10 doesn't
Am 21.07.2015 um 02:10 schrieb Ilia Mirkin:
On Mon, Jul 20, 2015 at 7:53 PM, Roland Scheidegger srol...@vmware.com
wrote:
Am 20.07.2015 um 22:46 schrieb Ilia Mirkin:
On Mon, Jul 20, 2015 at 3:53 PM, Brian Paul bri...@vmware.com wrote:
On 07/20/2015 01:19 PM, Ilia Mirkin wrote:
On Mon, Jul
Am 20.07.2015 um 22:46 schrieb Ilia Mirkin:
On Mon, Jul 20, 2015 at 3:53 PM, Brian Paul bri...@vmware.com wrote:
On 07/20/2015 01:19 PM, Ilia Mirkin wrote:
On Mon, Jul 20, 2015 at 2:44 PM, Ilia Mirkin imir...@alum.mit.edu wrote:
Hi Brian,
You marked off ARB_copy_image (and
Am 25.08.2015 um 01:08 schrieb Ilia Mirkin:
On Mon, Aug 24, 2015 at 7:04 PM, Brian Paul bri...@vmware.com wrote:
Mesa and gallium don't have a complete set of matching 3-component
texture formats. For example, 8-bit sRGB unorm. To fully support
the GL_ARB_copy_image extension we need to have
Thanks for fixing this up.
Reviewed-by: Roland Scheidegger <srol...@vmware.com>
Am 21.10.2015 um 18:25 schrieb Jose Fonseca:
> This should prevent disparity between features Mesa and LLVM
> believe are supported by the CPU.
>
> http://lists.freedesktop.org/archives/mes
Am 29.10.2015 um 12:27 schrieb Oded Gabbay:
> Hi Roland, Jose
>
> I wanted to bring a problem I found to your attention, and discuss
> about it and ways to solve it.
>
> I'm working on regressions of piglit gpu.py between x86-64 and
> ppc64le, when running with llvmpipe.
>
> One of the
Am 29.10.2015 um 20:10 schrieb Oded Gabbay:
> On Thu, Oct 29, 2015 at 9:02 PM, Ilia Mirkin wrote:
>> On Thu, Oct 29, 2015 at 2:44 PM, Oded Gabbay wrote:
>>> However, I would hate to keep the situation as is, meaning the test
>>> passes on x86-64 and
Am 29.10.2015 um 19:44 schrieb Oded Gabbay:
> On Thu, Oct 29, 2015 at 5:31 PM, Roland Scheidegger <srol...@vmware.com>
> wrote:
>> Am 29.10.2015 um 12:27 schrieb Oded Gabbay:
>>> Hi Roland, Jose
>>>
>>> I wanted to bring a problem I found to your
Am 29.10.2015 um 20:33 schrieb Oded Gabbay:
> On Thu, Oct 29, 2015 at 9:25 PM, Roland Scheidegger <srol...@vmware.com>
> wrote:
>> Am 29.10.2015 um 20:10 schrieb Oded Gabbay:
>>> On Thu, Oct 29, 2015 at 9:02 PM, Ilia Mirkin <imir...@alum.mit.edu> wrote:
>>&
Am 26.10.2015 um 19:31 schrieb Ilia Mirkin:
> On Mon, Oct 26, 2015 at 2:15 PM, Neil Roberts wrote:
>> Nanley Chery writes:
>>
>>> + /* Throw an INVALID_OPERATION error if the target is
>>> + * TEXTURE_CUBE_MAP_ARRAY and the format is not
Am 28.10.2015 um 20:48 schrieb Ian Romanick:
> On 10/28/2015 09:52 AM, Nanley Chery wrote:
>> From: Nanley Chery
>>
>> The ASTC spec forbids other compressed formats from being used against
>> the targets: TEXTURE_CUBE_MAP_ARRAY and TEXTURE_3D. Because other
>
> I
Am 26.10.2015 um 10:02 schrieb Jose Fonseca:
> On 23/10/15 22:26, srol...@vmware.com wrote:
>> From: Roland Scheidegger <srol...@vmware.com>
>>
>> f16c intrinsic can only be emitted when AVX is used. So when we
>> disable AVX
>> due to forcing 128bit
Am 26.10.2015 um 16:33 schrieb Jose Fonseca:
> On 26/10/15 14:58, Roland Scheidegger wrote:
>> Am 26.10.2015 um 10:02 schrieb Jose Fonseca:
>>> On 23/10/15 22:26, srol...@vmware.com wrote:
>>>> From: Roland Scheidegger <srol...@vmware.com>
>>>>
Ahhh legacy functionality from hell...
For the series:
Reviewed-by: Roland Scheidegger <srol...@vmware.com>
I'm wondering if it would be possible to omit the rasterPos execution
completely, by incorporating it into drawPixels etc. at least in the
hopefully common case there's no state c
Am 22.10.2015 um 00:41 schrieb Rowley, Timothy O:
>
>> On Oct 20, 2015, at 2:03 PM, Roland Scheidegger <srol...@vmware.com> wrote:
>>
>> Certainly looks interesting...
>> From a high level point of view, seems quite similar to llvmpipe (both
>> til
Oh and it probably should go to stable.
Roland
Am 21.10.2015 um 18:55 schrieb Roland Scheidegger:
> Thanks for fixing this up.
>
> Reviewed-by: Roland Scheidegger <srol...@vmware.com>
>
> Am 21.10.2015 um 18:25 schrieb Jose Fonseca:
>> This should prevent disp
Am 22.10.2015 um 18:27 schrieb Alex Deucher:
> The section for UVD 2 and older was not updated
> when HEVC support was added. Reported by Kano
> on irc.
>
> Signed-off-by: Alex Deucher
> Cc: mesa-sta...@lists.freedesktop.org
> ---
>
1001 - 1100 of 2156 matches
Mail list logo